improvement-analyst

improvement-analyst is an agent for Claude Code from gertsylvest/meta-team. It costs 31 tokens per session (1,067 once invoked), scanned A, original, MIT.

A review agent that examines Claude Code session transcripts—records of an agent's work—to find inefficient patterns and suggest improvements.

In plain words
What is it for?
Use it at sprint ends or checkpoints to create structured suggestions in a project-specific self-improvement log.
Why use it?
It helps reveal wasted steps, errors, and repeated loops that may be hard to notice while working.

Agent for Claude Code

Written for Claude Code: a Claude Code subagent (agents/*.md). Also seen: model in frontmatter; reads .claude/ paths; mentions subagents.

Not installable: its command points at a path on the author’s own machine, so it runs nowhere else. The line is /Users/name/dev/my-project.

Good fit Use it at sprint ends or checkpoints to create structured suggestions in a project-specific self-improvement log.

Compare 6 agents from other repositories ↓
Install

Getting it into your agent

There is no command for this one: it runs only inside a plugin, and the catalogue could not identify which plugin ships it. The source is linked below.

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for improvement-analyst

README.md
[![agentmods](https://agentmods.dev/badge/agents/gertsylvest/meta-team/improvement-analyst/github.svg)](https://agentmods.dev/agents/gertsylvest/meta-team/improvement-analyst)
Your own site
<a href="https://agentmods.dev/agents/gertsylvest/meta-team/improvement-analyst"><img src="https://agentmods.dev/badge/agents/gertsylvest/meta-team/improvement-analyst/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for improvement-analyst

Your own site · 80×15
<a href="https://agentmods.dev/agents/gertsylvest/meta-team/improvement-analyst"><img src="https://agentmods.dev/badge/agents/gertsylvest/meta-team/improvement-analyst.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 31 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,067 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00031 $0.01067
Opus 5 $0.00015 $0.00534
Sonnet 5 $0.00006 $0.00213
Haiku 4.5 $0.00003 $0.00107

Measured 9d ago against content hash 4b18114941d3, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

improvement-analyst scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

library/agents/improvement-analyst.md · 107 lines

How it starts

The opening of the file, as written. The whole thing — 107 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Role

You are the Improvement Analyst. You analyze Claude Code session transcripts to detect inefficiencies, write structured optimization suggestions to the project's self-improvement log, and signal findings back to the orchestrator.

You are invoked by the orchestrator at sprint end, or at any meaningful checkpoint during a working session.

Inputs

You receive from the orchestrator:

  • The current project root path (e.g. /Users/name/dev/my-project)
  • Optionally: a sprint label or time window to scope the analysis (e.g. "sprint-3", or "last 2 hours")

Step 1 — Locate session transcripts

Claude Code stores transcripts as JSONL files at:

~/.claude/projects/<encoded-project-path>/*.jsonl

The encoded project path replaces / with - in the absolute project path (e.g. /Users/gts/dev/my-project-Users-gts-dev-my-project).

Find all .jsonl files under the matching folder, including subagent sidechains at <session-uuid>/subagents/agent-*.jsonl (these carry isSidechain: true). This is critical: the top-level orchestrator transcript records only ~15% of tool activity — the bulk of tool calls, errors, and loops happen inside subagents. Scanning only the top-level sessions can undercount errors by an order of magnitude. Pass the project root to the log-pattern-scan skill (Step 2), which recurses into sidechains for you; do not hand-roll a top-level-only glob. If a time window was given, filter by file modification time using Bash (ls -lt).

Prefer the log-pattern-scan skill (run via bash, which is pre-authorized) and Glob/Read for inspection. Avoid ad-hoc compound or multi-line Bash (pipes, redirects, for loops) for scanning — those commands fall outside the standard allowlist prefixes and trigger permission prompts, which adds friction and can stall the analysis if a prompt is declined.

Step 2 — Scan transcripts for patterns

For each JSONL file, invoke the log-pattern-scan skill:

log-pattern-scan <jsonl-file-path> all

Read the full file on GitHub · 107 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 9d ago First seen · 107 lines · 31 tokens per session scan A 4b18114941d3

Subscribe to this mod's changes

improvement-analyst is an agent published in the GitHub repository gertsylvest/meta-team (5 stars, last pushed 1mo ago), licensed MIT. It adds 31 tokens to every session and 1,067 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.